J. Am. Chem. Soc. 1996, 118, 12457-12458
12457
Use of 13C-13C NOE for the Assignment of NMR Lines of Larger Labeled Proteins at Larger Magnetic Fields Mark W. F. Fischer,†,§ Lei Zeng,† and Erik R. P. Zuiderweg*,†,‡ Biophysics Research DiVision, Department of Biological Chemistry, and Department of Physics The UniVersity of Michigan 930 N. UniVersity AVenue, Ann Arbor, Michigan 48109 ReceiVed June 28, 1996 The methodology for the determination of solution structures by NMR has been optimized for (labeled) biomolecules of molecular weights of 20 kDa and less.1,2 At risk of becoming less useful for the study of molecules larger than that or molecular complexes are the triple-resonance2 and especially HCCH3 experiments that are at the very core of the current resonance assignment protocol. The 2D, 3D, and 4D HCCHCOSY and HCCH-TOCSY experiments that correlate the side chain resonances transfer coherence over a 1JCC coupling of ∼34 Hz and become insensitive when the 13C line widths approach this value, as for 30 kDa proteins: one computes efficiencies of only 14% for a 13 kDa, 10% for a 20 kDa, and 8% for a 30 kDa protein.4,5 Here, we demonstrate that magnetization transfer using the homonuclear dipolar interaction (NOE) between adjacent 13C nuclei should be considered to complement or replace scalar transfer in larger proteins. Figure 1A shows a calculation of the steady state {13CR}-13Cβ NOE difference in a multispin fragment as a function of the rotational correlation time τc and magnetic field B.6 One observes that the 13C-13C NOE difference is at the onset of the large molecule regime ((ωCτc)2 .1) for macromolecules with rotational correlation time τc of 10-8 s at all magnetic fields considered. In contrast to the scalar transfer, NOE efficiency improves significantly with increasing molecular size and magnetic field and is substantial in the large molecule limit, even though many relaxation mechanisms are present.7 Although maximum 13C-13C NOEs will not be attainable8 in the near future, sizable (-20%) NOE differences can currently be obtained at 17.6 T (750 MHz) for molecular weights around 30 kDa (log τc ) -7.65). Potentially more interesting is the analog of a 2D 1H-1H NOESY experiment (i.e., a 13C-13C NOESY experiment). * Author to whom correspondence should be addressed. † Biophysics Research Division. ‡ Department of Biological Chemistry. § Department of Physics. (1) Wu¨thrich, K. NMR of Proteins and Nucleic Acids; Wiley: New York, 1986. (2) For a review, see: Bax A.; Grzesiek, S. Acc. Chem. Res. 1993, 26, 131. (3) Fesik, S. W.; Eaton, H. L.; Olejniczak, E. T.; Zuiderweg, E. R. P.; McIntosh, L. P.; Dahlquist, F. W. J. Am. Chem. Soc. 1990, 112, 886. Bax, A.; Clore, G. M.; Gronenborn, A. M. J. Magn. Reson. 1990, 88, 425. Wang, H.; Zuiderweg, E. R. P. J. Biomol. NMR 1995, 5, 207. (4) We used the equation Emax = sin2(πJCCτmax) cos2(πJCCτmax) exp(-2πνC1/2τmax) with a value of 1 Hz/kDa for the line width of the 13C resonance (νC1/2) (25 °C); JCC is 34 Hz. (5) Side chain 13C line widths can be dramatically reduced by protein perdeuteration, offering one solution to the transfer efficiency problem in HCCH-COSY. See also: Kushlan, D. M.; LeMaster, D. M. J. Biomol. NMR 1993, 3, 701. Farmer, B. T., II; Venters, R. A. J. Am. Chem. Soc. 1995, 117, 4187. (6) Goldman, M. Quantum Description of High-Resolution NMR in Liquids; Oxford University Press: Oxford, 1988; p 248. (7) See Figure 1A in the following: Zeng, L.; Fischer, M. W. F.; Zuiderweg, E. R. P. J. Biomol. NMR 1996, 7, 157. (8) Perdeuteration of the amino acid side chains shifts the curves in Figure 1a approximately 0.5 log unit to the left (i.e., NOE difference approaches maximum for molecules with τc of 10 ns). Clearly, perdeuteration could be of help, but gains in efficiency are likely to be offset by forfeiting the 1H f 13C polarization transfer and, more importantly, loss in recycling efficiency.
S0002-7863(96)02200-7 CCC: $12.00
Figure 1. Computation of 13CR f 13Cβ NOE in the molecular fragment HN, NR, CR, HR, C′,Cβ, 2Hβ, Cγ, 2Hγ. (A) The steady state NOE difference for Cβ while saturating CR. We found that this 11-spin system could be adequately described by the linear three-spin equation Diff (see Supporting Information) NOERfβ = σβR/(Fβ - σ2βγ/Fγ), where 2 2 2 2 6 σCβ-Q ) [µ0/4π] (γCβγQp /10r Cβ-Q){6J(2ωC) - J(0)} and Fβ ) 2(σ⊥ 2 - σ|)2ω2C/15 J(ωC) + ∑HN,NR,CR,HR,C′,2Hβ,Cγ,2Hγ [µ0/4π](γCβ γ2QQ 2 6 p /10r Cβ-Q){3J(ωC) + 6J(ωC + ωQ) + J(ωC - ωQ)} where µo is the vacuum permittivity, γQ is the gyromagnetic ratio of Q, σ⊥ - σ| is the Cβ chemical shift anisotropy, ωc is the carbon angular Larmor frequency, r is the internuclear distance, and p is Planck’s constant over 2π.6 For simplicity, we assumed that Fγ ) Fβ. (B) The transient NOE (initial condition: inversion of CR) for the 13Cβ nucleus calculated through integration of a 3 × 3 relaxation matrix11 for CR, Cβ, and Cγ, where we considered 11 spins for the calculation of FR and Fβ and assumed Fγ ) Fβ. For both A and B, the Lipari-Szabo model12 was used to model the spectral density functions using an order parameter of 0.87 and a local correlation time of 10-11 s. A value of 32 ppm was used for the Cβ CSA anisotropy.13 Line styles in A represent, from right to left: long dashes, 11.75 T and 500 MHz 1H; solid line, 14.09 T and 600 MHz 1H; alternating dashes, 17.62 T and 750 MHz 1H; short dashes, 23.5 T and 1000 MHz 1H. Line styles in B represent: thin solid line, 20 kDa protein at 14.09 T and 600 MHz; alternating dashes, 20 kDa protein at 17.62 T and 750 MHz 1H; short dashes, 20 kDa protein at 23.5 T and 1000 MHz 1H; long dashes, 30 kDa protein at 17.62 T and 750 MHz 1H; thick solid line, 30 kDa protein at 23.5 T and 1000 MHz 1H.
Related spectra were recently demonstrated for measuring relaxation properties through 13CR-13CO NOE buildup.9 Figure 1B shows computed NOE buildup curves at different conditions; it follows that the 13C-13C NOE at 1 GHz will be three times as efficient as the 13C-13C scalar transfer for a similar size molecule. For 20 kDa molecules studied at 600 MHz, 13C13C NOE transfer and 13C-13C scalar transfer appear to be about equally efficient; therefore, we explored 13C-13C NOE experiments for the 19-21 kDa peptide-binding domains of the chaperone proteins Hsc and DnaK. We demonstrate here 2D H(CC)H and 3D (H)CCH versions of the following generalized 4D 1H-13C-13C-1H NOESY experiment: pH90 - {tH1 , 1/2JCH} C C C C - pC,H 90 - {t2 , 1/3JCH} - p90 - τM - p90 - {t3 , 1/3JCH} C,H H p90 - {1/2JCH} - t4 The 4D experiment samples the frequencies of two adjacent carbon nuclei through two half semiconstant time periods {tC2,3, 1/3JCH} around the 13C-13C NOE mixing time τm transfers the magnetization from and to (9) Cordier, F.; Brutscher, B.; Marion, D. J. Biomol. NMR 1996, 7, 163.
© 1996 American Chemical Society
12458 J. Am. Chem. Soc., Vol. 118, No. 49, 1996
Figure 2. (A) Results of a two-dimensional H(CC)H-NOESY experiment, using a 1.4 mM sample of the 19 kDa peptide-binding domain of the molecular chaperone Hsc70 at 25 °C in 2H2O. A 600 MHz Bruker AMX instrument with a 5 mm triple-resonance gradient probe was used. The 13C-13C NOE mixing time was 1.0 s. Total experimental time was 36 h. (B) 13C-13C NOESY and 13C-13C COSY cross sections at 5.2 ppm. The H(CC)H COSY experiment3 was recorded with the same number of scans and increments while using 13C-13C scalar transfer times of 9 ms. The assignments14 are for HR-CR-Cβ-Hβ connectivities except for the spin diffusive I90 HR-CR-Cβ-Cγ-H2 and HR-CR-CβCγ-H3 which are marked with asterisks.
the attached hydrogen nuclei using a half semiconstant time period {tH1 , 1/2JCH} and a half-(R)INEPT sequence {1/2JCH}.10 We emphasize that 13C magnetization is aligned along the (z axis during the 1 s 13C-13C NOE mixing period τm while any remaining transverse magnetization (very little after 1 s) is dephased by pulsed-field gradients during that period. Figure 2A shows a 2D H(CC)H-NOESY experiment recorded with this pulse sequence; it is seen that the spectrum contains many 13C13C NOE-mediated cross peaks. A cross section through the spectrum is shown in Figure 2B, together with an identical cross section through a 2D H(CC)H-COSY experiment. One indeed finds that 13C-13C dipolar and 13C-13C scalar transfer are about equally efficient for these conditions. We note that the spectra are complementary: narrow resonances are stronger in COSY, broader ones are stronger in NOESY. Figure 3 shows three 13C-13C planes from a 3D (H)CCH-NOESY experiment and gives the connectivity tracing for the residue Val24 of the 21 kDa protein DnaK. Panel B contains the one-bond 13C-13C NOEs 13Cβ (37 ppm) f 13CR (57 ppm), 13Cβ f 13Cγ1 (21 ppm), and 13Cβ f 13Cγ2 (19 ppm) for this residue. The reciprocal one-bond 13C-13C NOE connectivities are indicated in panels A and C, where the presence of a small amount of (10) The semiconstant time periods (Grzesiek, S.; Bax, A. J. Biomol. NMR 1993, 3, 185. Logan, T. M.; Olejniczak, E. T.; Xu, R. X.; Fesik, S. W. J. Biomol. NMR 1993, 3, 225) and reverse INEPT are {t,1/3JCH} ≡ a b c -τa-pC180-τb-pH180-τc-; {1/2JCH} ≡ -δ-pC,H 180 -δ- where τ - τ + τ ) 1/2JHC for the proton frequency labeling and 1/3JHC for the carbon frequency labeling. The length of the reverse-INEPT half-step 2δ is 1/2JHC. For more details see the Supporting Information. (11) Solomon, I. Phys. ReV. 1955, 99, 559. (12) Lipari, G.; Szabo, A. J. Am. Chem. Soc. 1982, 104, 4546. (13) Daragan, V. A. ; Mayo, K. H. Chem. Phys. Lett. 1993, 206, 30. (14) Resonance assignments for the peptide-binding domain of Hsc-70 were obtained in our lab by R. C. Morshauser. (15) Resonance assignments for the peptide-binding domain of DnaK are described: Wang, H. Ph.D. Thesis, The University of Michigan, 1995.
Communications to the Editor
Figure 3. Results of a three-dimensional (H)CCH-NOESY experiment using a 1.2 mM sample of the 21 kDa peptide-binding domain of the molecular chaperone DnaK at 25 °C in 2H2O. A 600 MHz Bruker AMX instrument with a 5 mm triple-resonance gradient probe was used. 13C13C planes (ω ,ω ) are shown at different 1H frequencies (ω ). A 13C2 3 4 13C NOE mixing time of 1.0 s was used. The data set was 44 × 80 × 1024 hypercomplex points and was recorded with 16 scans per increment using a 1.0 s relaxation delay. Total experimental time was 150 h. In A, B, and C, the 13CR, 13Cβ, 13Cγ1, and 13Cγ2 resonances of Val24 are traced15 at the proton frequencies 5.29, 1.59, and 0.73, respectively (the 1Hγ1 and 1Hγ2 resonances are nearly degenerate). See Supporting Information and text for details.
spin diffusion is apparent (e.g., 13Cγ1 f 13Cβ f 13CR and 13Cγ1 f 13Cβ f 13Cγ2 in panel C). Peak picks in the (13Cβ, 13CR, 1HR) region through the entire spectrum show that virtually all 13CR-13Cβ crosspeaks for the DnaK domain are present. It is therefore demonstrated that it is possible to obtain good 3D (H)CCH-NOESY spectra using a 5 mm sample of a 1.2 mM solution of a 21 kDa protein using a 600 MHz spectrometer. According to our calculations, shown in Figure 1B, one expects that the sensitivity of the HCCH-NOESY experiment will only increase when larger proteins will be studied at larger magnetic fields. The NOE method should thus be considered for tracing side chain connectivities and may possibly replace the current HCCH-COSY and HCCH-TOCSY experiments for the study of larger molecules. Acknowledgment. We thank Dr. G. C. Flynn (University of Oregon) for the samples of DnaK and Hsc70. We thank Dr. H. Wang and Mr. R. C. Morshauser for the assignments of the NMR spectra of DnaK and Hsc-70. This work was supported by NSF grant MCB9513355. Supporting Information Available: Derivation of an expression for the steady state NOE in a three-spin system; detailed description of the pulse sequence, gradient and phase cycles for the 4D HCCH NOESY experiment; experimental parameters for the 3D (H)CCH NOESY shown in Figure 3 (3 pages). See any current masthead page for ordering and Internet access instructions. JA962200+
